Post on 15-Mar-2020
transcript
19th International Conference on
TRANSPORT AND SEDIMENTATION OF SOLID PARTICLES
24-27 September 2019, Cape Town, South Africa
ISSN 0867-7964 ISBN 978-83-7717-323-7
OPTIMIZATION OF SAND TRAP AND SETTLER DESIGNS FOR
EFFICIENT DEPOSITION OF SUSPENDED SEDIMENT
Claudia Mc Leod, Ousmane Sawadogo and Gerrit Basson
Department of Civil Engineering, Stellenbosch University, Joubert Street, Stellenbosch,
South Africa, grbasson@sun.ac.za
Sediment traps are structures designed to allow the suspended sediment load, which enters river abstraction works or diversion canals, to deposit whilst relatively clean water passes through. The
deposited sediment is then removed by making use of gravitational flushing. The sediment traps
discussed in this paper are settlers and sand traps. The focus in this paper will be on optimizing the
design of settlers and sand traps to propose a set of design guidelines for use in South Africa. This is
done by investigating existing sediment traps within Southern Africa as well as numerically investigating some of the design properties of settlers and sand traps. A new concept of a sediment
trap is also numerically investigated as a possible design.
KEY WORDS: Sand trap, settler, settling velocity, settling length, numerical models, CFD
NOTATION
𝑤 Settling velocity (m/s)
𝑣 Kinematic viscosity (m2/s)
𝜌𝑠 Density of sediment (kg/m3)
𝜌 Density of water (kg/m3)
𝑔 Gravitational acceleration (m/s2)
𝑑 Particle diameter (m)
𝐿 Analytical settling length of a particle (m)
𝑉 Mean flow velocity (m/s)
ℎ Height at which a particle enters the trap (m)
𝑢∗ Shear velocity (m/s)
𝑅 Hydraulic radius (m)
1. INTRODUCTION
Sand traps and settlers are relatively large structures that require a fair amount of
building materials to be constructed, which makes them quite expensive. It is therefore
important to design an economically efficient and functional structure to remove sediment
and to ensure continuous supply of clean water to prevent pump or hydropower turbine
damage.
Currently, there are no specific guidelines to design sand traps or settlers in South
Africa and there have been reports that some of the existing sand traps and settlers
Claudia Mc Leod, Ousmane Sawadogo and Gerrit Basson
302
constructed are not working efficiently. The problems may lay in the sed iment
concentration imbalances between the intake and outlet due to inadequate lengths, flushing
ability, overall structural design or inefficient maintenance.
This paper investigates existing sediment traps in Southern Africa, as case studies, by
collecting design and field measurements for analysing sediment transport within the traps.
This is done in order to understand the problems (if any) and the performance of the
designed sediment trap structures. Some of the properties of existing settlers and san d traps
are numerically investigated by making use of ANSYS Fluent v18.1. The properties
include the overall structural design such as the dimensions, slope, cross -section and type
of inlet as well as the concentration intake.
A new concept of a settling basin developed at the Norwegian University of Science
and Technology by Støle (1997), is then numerically investigated. This concept is expected
to be more appropriate in projects where existing sediment exclusion facilities are
inefficient.
2. SEDIMENT SETTLING THEORY AND SEDIMENT TRAP DESIGNS
The use of settlers or sand traps prevents damage to pumps used in high lift pump
stations, hydropower turbines and sediment deposition in water conveyance systems. The
finer the sediment that can be trapped in a sediment trap, the less need there will be for
water treatment plants that uses flocculation methods to rid the water of fine sediment for
potable use. The trapped sediment within is flushed back to the river and therefore helps
in restoring a sediment mass balance within the river. Sediment that is abstracted at water
treatment plants cannot be reintroduced to the river as it contains toxic chemicals that can
affect aquatic life. The most important sediment properties regarding the design of settlers
and sand traps are the settling velocity and the settling length of a sediment particle.
2.1 SETTLING VELOCITIES OF SEDIMENT PARTICLES
Settling velocity is one of the main variables in the study of sediment transport for
understanding sediment suspension and deposition. Particle settling velocity is the speed
at which a particle will settle to the bottom of a body of water. Van Rijn (1989)
recommended to calculate the settling velocity of different sediment particles based on
their diameter size. The sediment velocities for each particle used in the numerical
simulations was calculated by making use of the following equations.
𝑤 =𝑣
18 (
𝜌𝑠
𝜌− 1) 𝑔𝑑2 for 𝑑 ≤ 0.1 mm
𝑤 = 10 𝑣
𝑑 [(1 + 0.01 (
𝜌𝑠
𝜌− 1)
𝑔𝑑3
𝑣2)0.5
− 1] for 0.1 < 𝑑 < 1.0 mm
𝑤 = 1.1 (𝜌𝑠
𝜌− 1) 𝑔𝑑
0.5
for 𝑑 ≥ 1.0 mm
Optimization of sand trap and settler designs for efficient deposition of suspended sediment
303
2.2 ANALYTICAL SETTLING LENGTH
The actual settling length of a specific sediment particle within a trap depends on the
height at which the particle enters and the flow velocity. In designing a sediment trap, one
must consider how long the trap must be to deposit all sediment down to a desired size.
Bouvard (1992) recommends using the following empirical formulas to determine the
length of the sediment trap. It is recommended that the length of the sediment trap is
extended by 10 to 20% in order to compensate for the excessive turbulence within the
transition zone at the inlet of the trap.
𝐿 = 𝑉 𝑥ℎ
𝑤−𝑢∗ with 𝑢∗ =
4.2𝑉
100𝑥
1
𝑅16⁄
2.3 SETTLER AND SAND TRAP DESIGN
A sediment trap is primarily designed according to the maximum size (𝑑𝑚𝑎𝑥 ) of the
sediment it needs to convey and the minimum diameter (𝑑𝑜) of the sediment that has to be
removed. Attention is centred on the minimum sediment diameter and the velocity (𝑉𝑑𝑜 )
required for deposition, as well as the minimum length required for the deposition of the
sediment.
Settlers are designed for a low discharge and flow velocity (0.1 to 0.2 m/s) to be able
to settle sediment particles larger than 0.3 mm if they have sufficient length. A settler is
flushed periodically and the particle size that can be removed is based on its slope. Sand
traps can handle a high discharge and therefore higher flow velocities (0.2 - 0.5 m/s) which
makes it difficult for these traps to deposit particles smaller than 0.3 mm. Sand traps are
flushed continuously at the outlet or distributed sediment scour holes. Optimisation of the
geometry of the cross-section of a canal is an important factor to consider in reducing the
costs of excavation and lining. A rectangular cross -section has the best hydraulic efficiency
if its water depth is half of the channel width. A trapezoidal cross -section is most
economical when the top width is double the length of one sloping side. Figure 1 shows
the cross-sectional specifications for the efficient design of a rectangular or trapezoidal
channel. These specifications were used to design the cross -sections of the rectangular and
trapezoidal settler models which is then numerically tested.
Figure 1. Cross-sectional specifications for most efficient cross-sections
Claudia Mc Leod, Ousmane Sawadogo and Gerrit Basson
304
The slope of the settler is designed according to the maximum size of sediment to be settled
out and flushed and to help with the deposition of fine sediment. Basson (2015) suggested
that the slope of a typical settler should be between 2 – 3%. There are no specific guidelines
for the inlet conditions of sediment traps, only that when designing one should minimize
the turbulence and velocity at the inlet transition zone.
3. CASE STUDIES: TIENFONTEIN SETTLER AND LUSIP SAND
TRAP
The Tienfontein settler is located near the Caledon River in the Free State Province of
South Africa. It has a length of 92 m, a width of 2.5 m and has been constructed with a
slope of 0.9% to allow for sediment deposition and flushing. The settler consists of three
operational canals and one standby canal which are all constructed in parallel. The flow
through each canal is 0.6 m³/s. The main objective of the Tienfontein set tler is to remove
sediment coarser than 0.3 mm. Field measurements have shown that the settler effectively
settles sediment coarser than 0.14 mm due to its sufficient length.
The Lower uSuthu Smallholder irrigation plant (LUSIP) sand trap is located near the
uSuthu River in eSwatini. It has a length of 70 m and a width of 8 m. Its effective settling
length is only 35 m due to turbulence caused by the upstream Avio gate. The sand trap
has an inlet discharge of 15.5 m3/s and a distributed scour discharge amounting to 2 m3/s.
The main objective of the LUSIP sand trap is to remove sediment coarser than 1 mm. The
LUSIP sand trap has reportedly not been functioning correctly. According to field
measurements done, sediment coarser than 1 mm escapes the trap and debris entering the
trap causes blockage of the scour holes. This means that both the sand trap and feeder canal
require regular manual cleaning.
The proposed solutions to this problem are to reduce the turbulence downstream of the
gate by installing baffle plates. This will increase the effective settling length. Another
solution is to construct another sand trap in parallel to reduce the flow and velocity within
the trap or to insert fine screens upstream of the sand trap to prevent debris from entering
and blocking the scour holes. Figure 2 below shows the LUSIP sand trap on the left and
the Tienfontein settler on the right.
Figure 2. The LUSIP sand trap (LHS) and the Tienfontein settler (RHS)
Optimization of sand trap and settler designs for efficient deposition of suspended sediment
305
4. NUMERICAL INVESTIGATION OF DESIGN PROPERTIES
Fully three-dimensional numerical models coupled in terms of flow field and sediment
transport are investigated by making use of ANSYS FLUENT v18.1. This Computational
Fluid Dynamics (CFD) package is equipped with a special feature named User Defined
Function (UDF) which allows the user to input their own functions in the simulations. The
concentration and settling velocities of different particle sizes was introduced to the model
by making use of the UDF and customised code. The numerical model validation is
discussed in Sawadogo (2015).
4.1 DESIGN PARAMETERS TESTED
The two basic settler models are tested; the first one has a rectangular cross -section and
the second one has a trapezoidal cross -section. Both models are 100 m in length and does
not have a longitudinal slope. Both these models are used to investigate the length required
to settle sediment diameters of 0.1, 0.2 and 0.3 mm with a settling velocity of 0.09, 0.026
and 0.044 m/s respectively.
For each sediment particle a simulation was carried out with the following varying
parameters: a sediment concentration (c) inlet of 1 and 10 kg/m3, heights of concentration
entering at the full depth and through the top third of the canal, and for inlet flow velocities
(v) of 0.1 and 0.2 m/s.
The effect of a 3% positive and 3% negative slope was then tested on both models again
with the sediment concentrations and inlet velocities varying, but a particle size o f 0.1 mm
and the height of concentration entering at full depth remained constant. The numerical
simulation results of the settling lengths were then compared to analytical calculated
settling lengths by making use of the equation mentioned in Section 2.2.
4.2 RESULTS OBTAINED
The analytical settling lengths were calculated by making use of the equations given to
calculate the settling velocities of sediment particles of a certain diameter for a velocity of
0.1 m/s and 0.2 m/s and a constant inlet height and hydraulic radius at both rectangular and
trapezoidal cross-sections.
The mesh used in the numerical model simulations are made out of tetrahedrons with
a maximum size of 0.2 m. The quality of the mesh is 0.8, which is very good. The velocity
was specified as an inlet condition and the concentration and settling velocity was
introduced to the model by making use of the UDF and code. The settling length was
determined after each simulation by looking at the concentration profiles taken in the
middle throughout the length of the settler. The particles settle out when the concentration
profile becomes constant, defining the settling length of the particle. The following tables
(Table 1 to Table 3) provide the numerical settling length results obtained from each
simulation.
Claudia Mc Leod, Ousmane Sawadogo and Gerrit Basson
306
Table 1
Rectangular settler model without a slope - Numerical simulated settling lengths
V = 0.1 m/s V = 0.2 m/s
Inlet height
Sediment diam. (mm) 0.1 0.2 0.3 0.1 0.2 0.3
Settling velocity (m/s) 0.009 0.026 0.044 0.009 0.026 0.044
Analytical set. L (m) 23.4 4.8 2.6 459.0 12.0 6.0
Whole_depth c = 1 kg/m3 (m) 25 9 5 *90 20 11
c = 10 kg/m3 (m) 25 9 5 *90 20 11
Top_only c = 1 kg/m3 (m) 25 9 5 *90 20 10
c = 10 kg/m3 (m) 25 9 5 *90 20 10
Note: * does not settle 100% effectively, there is still some sediment in suspension
Table 2
Trapezoidal settler model without a slope - Numerical simulated settling lengths
V = 0.1 m/s V = 0.2 m/s
Inlet height
Sediment diam. (mm) 0.1 0.2 0.3 0.1 0.2 0.3
Settling velocity (m/s) 0.009 0.026 0.044 0.009 0.026 0.044
Analytical set. L (m) 23.4 4.8 2.6 459.0 12.0 6.0
Whole_depth c = 1 kg/m3 (m) 25 9 5 *90 20 10
c = 10 kg/m3 (m) 25 9 5 *90 20 10
Top_only c = 1 kg/m3 (m) 25 9 5 *90 20 10
c = 10 kg/m3 (m) 25 10 5 *90 20 11
From the results in Table 1 and Table 2, it is evident that the concentration inlet quantity
(1 and 10 kg/m3) and inlet position (whole depth and top only) does not affect the settling
lengths for the different particles. The numerical settling lengths for both rectangular and
trapezoidal models are exactly the same, meaning that the cross -section does not play a
role in the numerical settling lengths of the particles. The analytical settling lengths
underestimates the numerical lengths by almost 50% for both the rectangular and
trapezoidal models. It is found that the settling length increases with approximately 100%
with an increase in velocity from 0.1 m/s to 0.2 m/s. It is evident that a flow velocity of 0.2
m/s cannot effectively deposit a sediment particle of 0.1 mm. The critical velocity for a 0.1
m particle is 0.1 m/s.
Optimization of sand trap and settler designs for efficient deposition of suspended sediment
307
Table 3
Rectangular and Trapezoidal settler model with a slope - Numerical model simulated settling
lengths
Note: * does not settle 100% effectively, there are still some sediment in suspension
- does not settle, particles escape settler
From the results in Table 3 it is seen that, for an inlet velocity of 0.1 m/s and 0.2 m/s,
a slope of +3% shortens the settling length of the 0.1 mm particle whereas for a slope of -
3%, the particles escape the settler through the outlet and does not settle at all. This is due
to fact that for a positively sloped settler, the cross -sectional area will increase going
downstream of the inlet, which will in turn cause the velocity to decrease and therefore the
particle will deposit over a shorter length. For a settler with a negative slope, the cross -
section will decrease, which will cause an increase in velocity. The velocity is higher than
the critical velocity of the sediment particle, which means it will be carried in suspension
and not deposit within the trap. It is therefore recommended to have a positively sloped
settler of +3%, to shorten the settling lengths of 0.1 mm particles and to effectively deposit
within the settler.
5. NUMERICAL INVESTIGATION OF A NEW SEDIMENT TRAP
A new concept has been developed at the Norwegian University of Science and
Technology (NTNU), by Dr. Støle in 1997. This concept is known as the “split and settle”
concept which directly refers to dividing the flow in a sand trap into sediment-free and
sediment-laden water and then removing the sediment from the water. As sediment-laden
water flows within a channel, the suspended sediment concentration increases near the
bottom. The split and settle concept then take advantage of the variation in sediment
concentration over the depth of flow by dividing the flow into an upper and lower part.
Figure 3 shows a side view illustration of the split and settle concept shaped by Støle.
Instead of settling all the specific suspended sediment in one operation, the flow is
divided into two or more channels and the same process is repeated until the water is of
satisfactory quality (Støle, 1993). The concept is said to be used for both pressurised and
gravitational flow conditions.
Rectangular Trapezoidal
Inlet velocity
Inlet height
Sediment diam. (mm) 0.1 0.1
Settling velocity (m/s) 0.009 0.009
V = 0.1 m/s
Whole_depth
No slope (m) 25 25
+3% slope (m) 15 15
-3% slope (m) - -
V = 0.2 m/s
Whole_depth
No slope (m) *80 *90
+3% slope (m) 50 50
-3% slope (m) - -
Claudia Mc Leod, Ousmane Sawadogo and Gerrit Basson
308
Figure 3. Side view illustration of the split and settle concept (Adapted from: (Støle, 1993))
5.1 SET-UP OF NUMERICAL MODEL
The numerical settling length obtained from the previous rectangular model CFD
simulations showed that the settling length of a particle with a diameter of 0.1 mm is
approximately 25 m. Therefore, the split and settle model was estimated to be 30 m long
with a split at 25 m. The split plate is situated 450 mm from the bottom and is 50 mm thick.
A velocity of 0.1 m/s was set at the inlet of the model which results in an inlet discharge
of 0.2 m3/s. The inlet concentration of sediment particles of size 0.1 mm was set at 1 kg/m3.
5.2 RESULTS OF NUMERICAL MODEL
The velocity vectors seen from the side of the canal and the velocity profiles throughout
the length of the model can be seen in Figure 4. The velocity magnitude vectors in the
model shows that it is constant at approximately 0.1 m/s throughout the length of the canal
and decreases underneath the split and increases at the top of the split. The decreased
velocities underneath the split is advantageous for the quicker deposition of sediment
particles.
Figure 4. Velocity profiles over depth along the length of the model
Optimization of sand trap and settler designs for efficient deposition of suspended sediment
309
The suspended sediment concentration map seen from the top of the canal and
concentration profiles over depth along the length of the model can be seen in Figure 5 and
Figure 6 respectively. From the concentration profiles along the depth of the canal it is
evident that sediment particles continue to deposit underneath the split and no sediment
concentration is observed above the split. From these results, it seems that the new model
is working as it should, but more numerical and physical model tests are recommended to
see how effective the design is and if it can be optimized.
Figure 5. Concentration map seen from the top of the canal (kg/m3)
Figure 6. Concentration profiles over depth along the length of the model
6. CONCLUSIONS
From the numerical investigation of the design properties it was found that the
numerical settling lengths for both rectangular and trapezoidal models for the same flow
velocities are exactly the same, meaning that the cross -section does not play a role in the
numerical settling lengths of the particles. The analytical settling lengths underestimates
the numerical lengths by almost 50% for both the rectangular and trapezoidal models. The
analytical settling length does not take the transition zone at the inlet into account. It is
found that the settling length increases with approximately 50% with an increase in
velocity from 0.1 m/s to 0.2 m/s. It is evident that a flow velocity of 0.2 m/s cannot
effectively deposit a sediment particle of 0.1 mm, but a settler with a +3% slope can deposit
the particle. This coincides with the case study results for the Tienfontein settler where
Claudia Mc Leod, Ousmane Sawadogo and Gerrit Basson
310
field measurements have shown that it effectively settles sediment coarser than 0.14 mm
due to its sufficient length and +0.9% slope. The new split and settle model seems to be
working as it should, but more numerical and physical model tests are recommended to
see how effective the design is. Although the length of the trap is shortened which makes
construction more economical, one should also consider the amount of water used for
continuous flushing of sediment in the model and if it is feasible.
ACKNOWLEDGEMENTS
The authors would like to thank the SA Water Research Commission (WRC) for
funding the study.
REFERENCES
1. Basson, G., 2015. River abstraction works and pipeline design. Sand trap design considerations,
23 September, pp. 212-226.
2. Bouvard, M., 1992. Mobile barrages and intakes on sediment transporting rivers. 2nd ed. Rotterdam: A.A Balkema.
3. Sawadogo, O., 2015. Coupled fully three-dimensional mathematical modelling of sediment
deposition and erosion in reservoirs, Stellenbosch: Stellenbosch University
4. Støle, H., 1993. Withdrawal of water from Himalayan rivers: sediment control at intakes.
Norway, Patent No. Ph. D. thesis. 5. van Rijn, L. C. (1984) ‘Sediment Transport, Part Iii : Bed Forms And Alluvial Roughness’,
Journal of Hydraulic Engineering, 110(12), pp. 1733–1754.